Re: Search Results In Error?

[Joe Shaw <joeshaw novell com>]

Hi,

On Mon, 2006-12-04 at 07:15 -0500, Rick Friedman wrote:
> Yesterday, I was doing a search with the search term being, "Maine". I
> was looking for all references to the state of Maine. Well, the search
> returned them fine. However, it also returned everything (web sites,
> files, etc) which contained the word, "main".
> 
> Shouldn't Beagle simply give me results for the search term I entered
> (Maine) and nothing else??

No, this is a feature called "stemming".  Maine stems to main, bake
stems to bak, country stems to countr.  Normally it's quite useful, as a
search for "baking" will match "bake" and "countries" will match
"country", etc.  You're hitting one of the unfortunately side effects of
that.

I believe the Snowball stemmers might be able to handle exceptional
cases like Maine, but I'm not sure and we're not using them presently.
I'd like to switch over to them as a result of the automatic language
detection work that is being done.

Joe